A Study in Ventricular–Ventricular Interaction
Single Right Ventricles Compared With Systemic Right Ventricles in a Dual-Chamber Circulation
Background Ventricular–ventricular interaction is known to occur in the normal human heart. To determine whether it plays a role in the function of single right ventricles, systemic right ventricles were compared with and without a left ventricle mechanically coupled to it.
Methods and Results A noninvasive magnetic resonance tagging technique (spatial modulation of magnetization [SPAMM]) that lays intersecting stripes down on the myocardium was used to examine 18 patients with systemic right ventricles: 7 with a single right ventricle who have undergone the Fontan procedure (age, 38.8±8.9 months) and 11 with transposition of the great arteries who have undergone an atrial inversion operation (age, 16.3±3.9 years). The motion of the intersection points was tracked through systole to determine regional twist and radial shortening. Shortening rates also were evaluated. Finite strain analysis was applied to the grid lines using Delaunay triangulation, and the two-dimensional strain tensor and principal E1 strains were derived for the various anatomic regions. Basal and apical short-axis planes through the ventricular wall were categorized into four distinct regions spaced equally around the circumference of the slice. We observed the following results. (1) Strain was greatest and heterogeneity of strain was least in patients with transposition of the great arteries who were status post atrial inversion operation (six of eight regions). Marked differences were noted in the distribution of strain within a given region, from endocardium to epicardium, and from atrioventricular valve to apical plane between patient subtypes and those with a normal left ventricle. (2) Contrary to the normal subject studied by the use of the same method, for both patient subtypes, there was counterclockwise twist in one region, clockwise twist in the posterior or inferior wall, and a transition zone of no twist at which the two regions of twist met. Normal human adult left ventricles studied in short-axis twist uniformly counterclockwise as viewed from apex to base. (3) Radial inward motion was greatest in the superior wall of both types of systemic right ventricle. The inferior walls of Fontan patients and the posterior (ie, septal) walls of patients with transposition of the great arteries, status post atrial inversion, moved paradoxically in systole. The shortening rate at the atrioventricular valve of patients with transposition of the great arteries, status post atrial inversion, was significantly lower than at the apex or in Fontan patients.
Conclusions Marked differences in regional wall motion and strain were demonstrated in systemic right ventricles, depending on whether a left ventricle was present to augment its function. Ventricular–ventricular interaction appears to play an important role in affecting the biomechanics of systemic right ventricles. These observations were markedly different from those in the normal systemic left ventricle. These techniques demonstrate tools with which we can begin to evaluate surgical outcomes using regional myocardial mechanics and may provide a clue to single right ventricle failure.
It has been demonstrated by numerous investigators that normal right ventricles (RVs) and left ventricles (LVs) do not act independent of each other and that ventricular–ventricular interaction (V-V) occurs.1 2 3 4 5 The mechanism of ventricular interdependence is the mechanical coupling of the ventricles, and it is demonstrated by a contribution of the LV to RV pressure generation (the reverse also is true4 ).
Patients with single RVs presently undergo a staged surgical reconstruction culminating in the Fontan procedure.6 7 8 9 Long-term success and viability of the single ventricle have been evaluated after surgery by the use of hemodynamic and anatomic parameters only.9 10 We have observed that a subgroup of these patients present with RV failure of unknown cause (eg, clinical congestive heart failure, poor systolic shortening, high end-diastolic pressure, significant tricuspid regurgitation, and so on). Because by definition single RVs do not have a second ventricle to augment function, the question arises of whether the lack of interaction with an LV contributes to failure in some single RVs.
We had an opportunity to study patients with systemic RVs with and without an associated LV to determine V-V interaction by comparing single RVs in patients who have undergone the Fontan procedure (Fontan patients)6 7 8 9 10 with those who have transposition of the great arteries (TGA) and have undergone an atrial inversion procedure (ie, Senning or Mustard operation) (status post [S/P] TGA patients).11 12 13 Both patient groups have postoperative intra-atrial baffles, making them even more comparable.
Using a noninvasive magnetic resonance tagging technique, we previously demonstrated that significant alterations exist in the myocardial mechanics across surgical stages of the Fontan procedure and between myocardial regions at each stage.14 15 In the present study, we evaluated regional myocardial strain and wall motion of the single RV after Fontan repair and S/P TGA by using a newly developed magnetic resonance imaging technique to determine regional myocardial twist, radial contraction, and strain. The normal human LV studied in our laboratory is used for reference.16
Regarding twist, as early as 1669, Richard Lower observed that cardiac motion was like the “wringing of a linen cloth to squeeze out the water.”17 Subsequent investigators have demonstrated and quantified LV twisting along its long axis, although there is debate about direction.17 18 19 20 21 22 23 24 25 26 27 28 29 We and others26 28 29 believe there is an initial LV counterclockwise twist (viewed apex to base) in early systole and that when the apex stops twisting, the base begins to untwist with wringing of the myocardium.
Strain, a unitless measure of deformation, is believed to be coupled with rotational dynamics17 19 and relates myocardial wall distortion during systole to end diastole (ie, the fractional change in dimension or size of the myocardium from the unstressed dimension that results from the application of a stress, which is defined as force divided by cross-sectional area18 30 ). It is a component of the regional tension–area diagram, which measures regional work, and has been shown to correlate with regional oxygen consumption in certain instances.31 Numerous studies have evaluated stress and strain in the normal LV,16 29 32 33 34 35 36 37 including the use of magnetic resonance tagging,16 29 37 38 39 40 which has recently been validated.38 These studies have suggested that due to the interplay between torsion and contraction, LV strain and shortening are distributed in a specific manner across the ventricular wall.
We hypothesize that single RV mechanics (twist, regional radial motion, and associated regional strain) differ from that of the RV of S/P TGA patients and from the normal LV because of a lack of V-V interaction. New image-acquisition techniques (magnetic resonance imaging and spatial modulation of magnetization [SPAMM]38 39 40 ) coupled with a broad-based image-analysis capability have made this in vivo evaluation possible.
We prospectively studied 18 patients with a systemic RV who were followed at The Children’s Hospital of Philadelphia between July 30, 1992, and August 30, 1993. Seven patients had a single RV (all had hypoplastic left heart syndrome) and had undergone the Fontan procedure.6 7 8 9 10 Eleven patients had TGA with intact ventricular septum and were S/P an atrial inversion procedure (nine Mustard procedures11 and two Senning procedures12 13 ). All patients were clinically well from a cardiovascular standpoint. No S/P TGA patients had tricuspid regurgitation or pulmonic stenosis. For Fontan patients, age ranged from 31 to 56 months (mean age, 38.8±8.9 months), and mean heart rate was 88±12 beats per minute. For S/P TGA patients, age ranged from 10 to 22 years (mean age, 16.3±3.9 years), and mean heart rate was 74±11 beats per minute. Mean time from operation was 15.6±4.2 months for Fontan patients and 11.3±3.9 years for S/P TGA patients. Patients had to be sufficiently stable to undergo a 1-hour magnetic resonance imaging scan while under sedation. Informed consent was obtained for all participants. The human investigations committee at The Children’s Hospital of Philadelphia approved the study protocol on February 4, 1992. No patient had exclusionary arrhythmias.
Magnetic Resonance Imaging
All Fontan patients and 4 of 10 S/P TGA patients were sedated before imaging. If less than 2 years old, the patient was administered either chloral hydrate (75 to 120 mg/kg PO) or Nembutal (2 to 6 mg/kg IV). If more than 2 years old, either Nembutal (4 mg/kg PO) and Demerol (3 mg/kg PO), Nembutal (2 to 6 mg/kg IV), or Ativan (5 mg PO) was administered. All patients were monitored with pulse oximetry, nasal end-tidal CO2, ECG, and direct visualization via television. All patients tolerated sedation without incident.
Studies were performed with a Siemens 1.5-T Magnetom. The scanning protocol was as follows. First, a stack of coronal localizers were acquired to locate the heart in the chest. T1-weighted transverse images that spanned the region of the heart were then acquired to evaluate cardiovascular anatomy and were used as a localizer for subsequent magnetic resonance tagging. The effective repetition time (TR) equaled RR interval (range, 350 to 800 milliseconds); echo time (TE) equaled 15 milliseconds; and the number of excitations was three.
Second, the method designed to standardize the RV short axis is shown in Fig 1⇓. The short-axis plane was defined as being perpendicular to the long-axis plane, and the long-axis plane was defined as being perpendicular to the atrioventricular valve plane and intersecting both the atrioventricular valve and the apex. Short-axis images were acquired by the use of the following localizers (Fig 1⇓). (1) Four to six T1-weighted images were obtained through the plane of the atrioventricular valve by the use of the transverse images as a localizer (one signal average). This ensured that there were images in the plane of the atrioventricular valve (Fig 1⇓, images I and II). (2) On an image that contained both atrioventricular valves or their remnants (in cases of atresia or stenosis), an imaginary estimated line was drawn between the two valves (images III and IV). Four to six T1-weighted images were obtained perpendicular to this imaginary line to provide an image of both the atrioventricular valve and the apex in a ventricular long-axis view (Fig 1⇓, image IV). (3) With the ventricular long-axis view (image V), an imaginary line was drawn that was the length of the ventricle through both atrioventricular valve and apex. This line was divided into thirds, and two images (one is one third of the way from the atrioventricular valve to the apex [designated “atrioventricular valve”] and one is two thirds of the way from the atrioventricular valve to the apex [designated “apex”]) were obtained orthogonal to this (image VI) to provide short-axis SPAMM images. End diastole was determined by the R wave on ECG.
Third, regarding myocardial tagging, the SPAMM sequence38 39 40 uses multiple radiofrequency pulses of 130° separated in time and a series of gradient radiofrequency pulses to produce saturated spins in two sets of parallel stripes that are perpendicular to each other (black on the image) followed by a standard gradient echo sequence dividing the wall into “cubes of magnetization” (Fig 2A⇓ and 2B⇓). Images 4 to 7 mm thick are acquired every 25 milliseconds for 12 phases from end diastole, and in plane tissue motion moves and distorts these cubes of magnetization. Tracking of this movement and distortion allows assessment of motion and deformation. TR equaled RR interval (range, 350 to 800 milliseconds); inversion time (TI) equaled 16 milliseconds; flip angle equaled 30°; the number of excitations was three; and the matrix size was 256×256. Grid lines were spaced to allow three or four lines to be laid down between endocardial and epicardial surfaces (ie, three or four rows of cubes; Fig 2A⇓ and 2B⇓). Tag thickness ranged from 1.5 to 2 mm.
Images were downloaded from the Magnetom onto a Sun SPARC 10 workstation (Sun Microsystems). vida (Volumetric Image Display and Analysis)41 was used to manipulate the images; this system uses a unix/x Windows graphic interface with a shared memory structure designed to hold multiple-volume image sets. Images were displayed with the use of a color scale that was linear over the dynamic range of the Sun video monitor. The window and level display parameters were held constant throughout the analysis of a complete cardiac cycle for a given patient, and the contrast and brightness controls of the Sun monitor always were set to their maximum gain.
To determine wall motion and strain, we tracked manually each SPAMM line intersection point (Fig 2A⇑, image I) through systole (typically for at least six phases) with a computer-based image with the use of a video cursor and mouse (Fig 2A⇑, image II). Next, the intersection points were connected automatically to form uniform, nonoverlapping triangles using the mathematical technique of Delaunay triangulation42 43 (Fig 2A⇑, image III). Using the centroid of each triangle (Fig 2A⇑, image IV) and wall motion software, we computed the regional wall motion across the short-axis plane of the ventricle (Fig 2A⇑, image V). Furthermore, complete two-dimensional finite strains (a measure of the amount of deformation experienced locally by the myocardium18 30 ) within each triangle were computed at the measured phases (Fig 2B⇑) with the use of strain software. The SPAMM technique has been validated for these measurements.38
Regional wall motion was determined by mathematically computing the centroid of all triangles during each phase of systole that was imaged. Twist and radial shortening were measured relative to the ventricular center of mass (Fig 2A⇑, image VI). The coordinates (x and y) of the centroid of the ventricular cavity were obtained at end diastole through computer-based user-interactive tracing of the endocardial border and then computation of the center based on the border location. With a global Cartesian coordinate system, the difference between the distance of the centroid of triangleA on the myocardial wall created by Delaunay triangulation42 43 to the centroid of the ventricular cavity at phase n and n+1 (ie, how far the muscle marked by the triangle was displaced away from the centroid of the ventricular cavity from phase n to phase n+1) is represented by the following equation:
where (xn, yn), and (xn+1, yn+1) are the coordinates of the centroid of triangleA at phase n and n+1, respectively; (xc, yc) are coordinates of the centroid of the ventricular cavity; and Pn and Pn+1 are vectors from the centroid of the cavity to the centroid of triangleA at phase n and n+1, respectively. The angle Θ (made by the two vectors drawn from the centroid of the ventricular cavity to the centroid of triangleA at phases n and n+1 (ie, twist around the centroid of the ventricular cavity) is represented by the following equation:
where Pn and Pn+1 are vectors (IPnI and IPn+1I are the lengths of those vectors) from the centroid of the cavity to the centroid of triangleA at phase n and n+1, respectively, and Pn · Pn+1 is the vector dot product. Wall motion data were displayed as depicted in Fig 2A⇑, image V. “Dots” represent end-diastolic triangle centroid location, and “tails” represent the subsequent motion. By convention, clockwise motion was negative and counterclockwise was positive, when viewing the heart apex to base. Parameters that were quantified were the magnitude and direction of net twisting; motion throughout systole (sampling at six to eight time points) also was observed graphically (Fig 2A⇑).
Radial motion was measured as the net inward motion of each triangle’s centroid toward the centroid of the ventricular cavity relative to the end-diastolic distance (measured in pixels of distance moved divided by initial radial length to normalize for heart size [Fig 2A⇑, images V and VI]) and was displayed to depict each triangle’s centroidal motion. By convention, radial motion inward was positive, and radial motion outward was negative.
Analysis necessitated dividing the myocardial wall into various standard regions (anterior, posterior, superior, and inferior walls).
As mentioned, finite strain analysis relates myocardial deformation during systole and uses end diastole as a reference (Fig 2B⇑).18 30 This parameter of cardiac mechanics is unitless and, in its simplest form in two dimensions, is defined as initial shape and size minus final shape and size divided by initial shape and size. From two-dimensional datasets, even the most complex finite deformation can be decomposed into two length changes and two associated angle changes. These measurements can easily be converted to finite strains of continuum mechanics (consisting of two normal strain vectors and two shear strain vectors). Based on previous work,36 two-dimensional finite strains were computed and are reported in the present study as their associated principal E1 strains.
Strain calculations used the two-dimensional strain tensor and the method of eigen system solutions to solve for principal strains, which are the maximum and minimum strains experienced by each triangle. Principal strain E1, reported in the present study, represents the most negative strain value (myocardial compression) (Fig 2B⇑). Strain in a given region at a given phase was obtained by averaging the strain of all the triangles within the region and reporting it as mean±SD. Our research focused on the maximum strain observed in the region throughout systole. An underlying assumption of our analysis is that cardiac muscle deformation is homogeneous within each triangle. The data were quantified and displayed in a color-coded format superimposed onto the anatomic images (Fig 2B⇑).
Because of regional wall motion abnormalities and because standardization of velocity of circumferential fiber shortening is lacking in the morphological right ventricle, fractional area shortening per unit time was used for shortening rate calculations. To determine systolic and end-diastolic cavity areas at the atrioventricular valve and apical short-axis levels, the endocardial borders at end diastole and in mid to late systole were traced manually (Fig 2A⇑, image VI) with the aid of a mouse. Computerized counting of the pixels enclosed by this border yielded the area; the following formula was used:
where ED is end-diastolic, area is measured in square millimeters, and time is measured in seconds. Time between images was ≈150 milliseconds.
Comparisons between two mean values were made with the use of the unpaired, two-way, Student’s t test and the Wilcoxon ranked sum test. Differences between various groups of patients and location (regional wall location, short-axis slice of either apex or base) were analyzed with the use of two-factor ANOVA with repeated measures used when appropriate. Comparison between multiple mean values within groups was made with the use of one-way ANOVA, with pairwise comparisons made with the use of Scheffé’s F test or Fisher’s protected least squares test. All measurements are given as mean±SD. Intraobserver variability was determined by replicate measures using the coefficient of variability. One trained observer performed all image-analysis steps. Significance was defined as P<.05. Statistical analysis was performed with a Macintosh II computer using statview ii version 1.03 software (Abacus Concepts).
To obtain the homogeneity of strain within the region, the coefficient of variation was used and derived as follows:
To compare endocardial strains with epicardial strains and to compare strains between atrioventricular valve and apical short-axis planes, the natural logarithm (ln) of the ratio of the endocardial to epicardial strain (eg, ln [endocardial strain/epicardial strain]) was used. This was advantageous in that significant differences implied significant differences in the geometric mean values of the two groups being compared.
Although the present study resulted in an extensive number of comparisons both within and between subjects, we have chosen to condense a group of related and statistically significant observations that demonstrate a pattern of altered regional wall motion and strain distribution for Fontan and S/P TGA patients that are not present in normal human hearts that we studied16 and that others studied, using similar methods.26 27 29 In each category (strain and wall motion), data are grouped according to short-axis level (atrioventricular valve and apical).
Fig 3A⇓ through 3C⇓ displays strain data in a color-coded format superimposed over the anatomic image. Fig 4A⇓ and 4B⇓ displays principle compressive strain (E1 multiplied by −1) for both Fontan and S/P TGA patients in the four anatomic quadrants (anterior, inferior, posterior, and superior walls) evaluated in two short-axis planes (one near the atrioventricular valve and one near the apex). Coefficient of variation for strain measurements was 6.2±1.6%. At the atrioventricular valve level (Fig 4A⇓), S/P TGA patients demonstrated greater compressive strain in the anterior and posterior walls (negative strain value furthest from zero) than Fontan patients (P<.05). Among Fontan patients, significantly less strain was found in the anterior wall than other wall regions (P<.05), whereas among S/P TGA patients, greater strain was found in the posterior wall (P<.05). At the apical level (Fig 4B⇓), S/P TGA patients had greater strain than Fontan patients in all regions except the posterior wall (P<.05).
Distribution of Strain
Heterogeneity of Strain Within a Given Anatomic Region
At the atrioventricular valve level (Fig 5A⇓), Fontan patients had the greater heterogeneity of strain in all regions except the superior wall (P<.05). Anterior and superior walls among S/P TGA patients and the anterior wall among Fontan patients displayed the greatest heterogeneity of strain (P<.05). At the apical level (Fig 5B⇓), Fontan patients had greater heterogeneity of strain than S/P TGA patients in only two regions (anterior and inferior), whereas S/P TGA patients had a greater heterogeneity of strain in the posterior wall (P<.05). It was the anterior and inferior walls among Fontan patients and the posterior wall among S/P TGA patients that had greater heterogeneity of strain than other regions (P<.05).
Endocardial–to–Epicardial Strain Ratio Within a Given Anatomic Region
Fig 5C⇑ and 5D⇑ displays the ln of the endocardial–to–epicardial strain ratio (a positive value implies endocardial>epicardial strain; a negative value implies endocardial<epicardial strain). At the atrioventricular valve level (Fig 5C⇑), Fontan patients differed from S/P TGA patients in all regions (P<.05). Among Fontan patients, inferior and posterior walls had positive values (ie, endocardial>epicardial strain as in the normal human LV16 ), whereas the anterior and superior walls had negative values. Among S/P TGA patients, the superior wall had a significantly positive value (P<.05), whereas all other regions were not significantly different from 0 (ie, endocardial=epicardial strain). At the apical level (Fig 5D⇑), S/P TGA patients demonstrated endocardial<epicardial strain (negative values, opposite of the normal LV pattern16 ) in all four regions, which differed significantly from Fontan patients, who had only two of four regions with that profile (more negative than S/P TGA patients at the inferior wall, less negative at the posterior wall) (P<.05). Among S/P TGA patients, the posterior wall had the most negative value of all regions (ie, endocardial<<epicardial strain) (P<.05).
Atrioventricular Valve–to–Apical Plane Strain Ratio (Distribution of Strain Along the Long Axis of the Ventricle)
Fig 5E⇑ displays the ln of the atrioventricular valve–to–apical plane strain ratio (a positive value implies atrioventricular valve>apical plane strain; a negative value implies atrioventricular valve<apical plane strain). S/P TGA patients differed significantly from Fontan patients in all four regions (P<.05). Among Fontan patients, all regions had positive values (ie, atrioventricular valve plane>apical plane strain, the pattern followed by normal LVs23 ). Among S/P TGA patients, the posterior wall had the highest positive value of all wall regions in either patient subtype, whereas the superior wall showed a significantly negative value (ie, atrioventricular valve plane<apical plane strain) (P<.05). Both anterior and inferior walls did not differ significantly from 0 (ie, atrioventricular valve plane=apical plane strain).
Fig 3A⇑ through 3C⇑ (left) depicts in graphic format the twisting motion in short axis of Fontan patients, S/P TGA patients, and the normal LV as studied in our laboratory, whereas Fig 6A⇓ and 6B⇓ quantitatively displays the data. Intraobserver variability was 5.6±1.9%. At the atrioventricular valve level (Fig 6A⇓), Fontan and S/P TGA patients twisted significantly differently in the inferior and posterior walls (S/P TGA patients had counterclockwise [ie, +] movement and Fontan patients had clockwise movement at the inferior wall, whereas at the posterior wall, S/P TGA patients had significantly more clockwise twist than Fontan patients [P<.05]). Among Fontan patients, both inferior and posterior walls twisted clockwise, whereas the superior wall did not twist and the anterior wall twisted counterclockwise. Among S/P TGA patients, only the posterior wall moved clockwise, whereas the other walls moved counterclockwise. At the apical level, however (Fig 6B⇓), it was only the posterior wall in both patient types that demonstrated twist in the clockwise direction, whereas all other walls moved counterclockwise (P<.05). The superior wall of Fontan patients demonstrated the most counterclockwise movement of all regions (P<.05).
Note the uniform counterclockwise motion of the normal LV (Fig 3A⇑) across both short-axis planes. Compare this motion with that of the single RV, S/P the Fontan procedure (Fig 3B⇑) and the systemic RV of S/P TGA patients (Fig 3C⇑). Note the similarity of twisting motion in both types of RVs, with regions of clockwise and counterclockwise twists that meet in a region of no twist that we call the “transition zone,” which had any one of three types of radial motion (inward, outward, or paradoxical).
Fig 3A⇑ through 3C⇑ displays in graphic format and Fig 6C⇑ and 6D⇑ displays in quantitative format the net radial motion along the short axis of the ventricle. Intraobserver variability was 5.4±1.7%. At the atrioventricular valve level (Fig 6C⇑), S/P TGA patients and Fontan patients had movement in opposite directions at the inferior and posterior walls (P<.05). Both patient types at this short-axis level had paradoxical systolic wall motion and differed only in the region in which it occurred (the posterior [septal wall] for S/P TGA patients and the inferior wall for Fontan patients). Both patient types had the greatest inward radial motion at the superior wall at both the atrioventricular valve and apical levels (P<.05). Similar to the atrioventricular valve level, at the apex (Fig 6D⇑) both patient types had paradoxical systolic wall motion, differing only in the region in which it occurred (the posterior [septal wall] for S/P TGA patients and inferior wall for Fontan patients), and the superior wall had the greatest inward radial motion.
Shortening rates differed significantly between Fontan and S/P TGA patients at the atrioventricular valve short-axis level (2.02±0.23 and 1.51±0.18 s−1, respectively, P<.05) but not at the apical short-axis level (1.99±0.12 and 2.22±0.34 s−1, respectively). Although there was no significant difference between atrioventricular valve and apical short-axis levels in the Fontan group, a significant difference did exist in the S/P TGA group (see above; P<.05).
V-V interaction1 2 3 4 5 is believed to be a result of a physical link between the two ventricles. Some have attributed this relationship to the interventricular septum, but others, noting that the myocardium constituting the ventricles forms an anatomic continuum around these chambers, have suggested that the free walls may also affect the contralateral ventricle independent of septal bulging.2 The “mechanical coupling of the ventricles” has been demonstrated in humans through the use of dP/dT of RV and LV pressure tracings.1 The RV was noted to affect LV pressure generation as well.4
We observed a number of patients with single RVs who have undergone the Fontan procedure to present with failing ventricles. To supplement hemodynamic and anatomic information, we evaluated regional wall mechanics in the hope of gaining a broader understanding of heart function and to determine the effects of V-V interaction. We previously evaluated the hearts of patients at various stages of Fontan reconstruction with the use of magnetic resonance tagging.14 15 44 Based on standard anatomic regions, we have demonstrated that distinct mechanical differences, based on strain measures, can be demonstrated between surgical subgroups.14 15 We have also shown that wall motion of the single ventricle is distinct and significantly different from that of the normal LV and that when the myocardium is divided according to patterns of motion, a region of significantly increased strain (the transition zone) is found.14
By defining the altered twist, shortening, and strain relations observed in the present study, it is our hope to have added to the groundwork needed to follow individual patients on a long-term basis and to begin to understand circumstances leading to and/or predicting failure. We further hope to define the role that V-V interaction, or the lack of, plays in the functioning of the single RV and whether V-V interaction is necessary for the long-term health of the heart.
In the Fontan procedure,6 7 8 9 10 all systemic venous return is channeled into the pulmonary arteries and bypasses the ventricle through direct anastomosis of superior vena(e) cava(e) to the pulmonary artery and through a hemicylindrical polytetrafluoroethylene baffle in the atria to direct inferior vena caval flow to the pulmonary artery. In the atrial inversion operation,11 12 13 a baffle is also placed in the atria to baffle systemic venous blood to the mitral valve and pulmonary venous blood to the tricuspid valve. Hoffman28 and Cristesue et al45 noted that atrial fibrillation and perhaps reduced atrial compliance or loss of atrial systole can cause an important ventricular afterload. The atrial baffles of both patient types that we studied may act similarly to cause an increased stress on the ventricle and may affect its long-term function. This may be why both types of systemic RVs differ functionally from the normal human LV.
Our analysis broke down ventricular biomechanical properties not only between Fontan and S/P TGA patients but also between regions within each patient subtype. Results from the present study indicate that strain in certain regions differed depending on whether a second ventricle was present to augment its function. Furthermore, strain was not distributed uniformly around the ventricular short axis in a given patient subtype, but rather certain walls appeared to deform to a greater degree than others, again, depending on whether an LV was present. Strain was greater in six of the eight regions studied (four regions at two short-axis levels) in S/P TGA patients than in Fontan patients and, although it did not reach statistical significance, was greater in a seventh. It is interesting to note that in S/P TGA patients, the posterior wall (ie, septal wall) is shared by both ventricles and is under the greatest strain. As a reference, in the normal human LV16 studied in our laboratory, the posterior free wall demonstrated the highest strain at the atrioventricular valve and apex levels. It may be that the geometric changes induced in the septum (ie, concave on the RV side and convex on the LV side, whereas the normal heart is the reverse) by such factors as fiber angle orientation and transmural pressure gradients play a role in this finding.
As noted, due to the interplay between torsion and contraction, normal LV stress and strain are distributed in a specific manner across the wall.29 32 33 34 35 36 37 Heterogeneity of strain is a measure of this nonuniformity. Within a given region, this distribution of strain depended on whether an LV was present to augment function. Five of eight regions demonstrated that not only were S/P TGA patients under greater strain but also the heterogeneity of this strain within a given region was less than that for Fontan patients, and in two of eight regions, there was no difference between patient subtypes. Also of note is that in general, regardless of whether the patient was S/P TGA or Fontan, the greater the strain, the more homogeneous was the strain within that region.
Other measures of distribution of strain in the myocardium (endocardial–to–epicardial and atrioventricular valve–to–apical plane strain ratios) also showed significant differences between patient types and from the normal human LV.16 In a number of regions, both S/P TGA and Fontan patients displayed a reversal of the endocardial–to–epicardial strain ratio (ie, endocardial strain<epicardial strain) as well as a reversal of atrioventricular valve–to–apical plane strain ratio (ie, atrioventricular valve strain<apical strain) of those of the normal human LV.16 There also were significant differences between S/P TGA and Fontan patients in all regions, demonstrating that V-V interaction affects not only absolute amounts of strain but also its distribution.
Twist and radial motion also were affected. Twisting on a microscopic level, is believed to be a combination of complex fiber architecture coupled with electrical activation sequences, allowing sarcomere shortening to be distributed in a specific manner throughout the thick wall of the normal LV17 18 19 20 21 22 23 24 25 26 27 (Fig 3C⇑; data from our laboratory). This torsion is believed to allow sarcomere stress, strain, and energy requirements to be minimized across the myocardium independent of loading conditions.18 19 20 Magnetic resonance tagging38 39 40 has been used successfully by many researchers to demonstrate the normal pattern of twist, shortening, and strain distribution.16 26 27 29 37
Twisting in both patient types was noted to be markedly different from normal at each slice level (Figs 3A⇑ through 3C⇑). Regions of both clockwise and counterclockwise twist were noted, with the abnormal clockwise motion being most notable in the posterior wall of both patient types and the inferior wall of Fontan patients. By necessity, these regions must meet in an area of no twist, which we call the transition zone. It is interesting that the regions of clockwise twist (not found in the normal LV) were also the same regions under the greatest strain in a given patient subtype (ie, the posterior wall of S/P TGA patients at the atrioventricular valve level and the posterior wall of Fontan patients at the apex). This abnormal twisting may imply an altered orderly reciprocal emptying and filling of the atrium and ventricle, possibly leading to inefficient energy use. This altered systemic RV motion may be a consequence of geometry altering foreign materials used to reconstruct the great vessels (affecting blood flow and velocity profiles), the mechanics of baffle placement within the atria affecting ventricular performance, or complications of cardiopulmonary bypass and deep hypothermic circulatory arrest.
Radial motion is also a measure of contraction and the ability of the heart to pump blood. It appears that depending on patient type, one wall moves paradoxically in systole (inferior wall in Fontan patients and the posterior wall in S/P TGA patients). Furthermore, radial contraction was not distributed uniformly around the short axis. At both the atrioventricular valve and the apex, regardless of whether the systemic RV had an LV attached, the superior wall underwent the largest radial inward motion. It is interesting that S/P TGA patients have their posterior wall (ie, septal wall) move paradoxically, as if the LV has not remodeled its fibers to convert from a systemic to a pulmonary pumping chamber. It is also interesting to note that the superior wall, which performed the greatest radial contraction in systemic RVs, also had the greatest counterclockwise twist, which implies that twist is coupled with shortening, consistent with published LV data.17
This difference from the normal motion of the LV myocardium (Fig 3C⇑; data from our laboratory) may be due to remodeling of fiber angles, extra cardiac tethering of the ventricle, altered electrical activation sequences, or ischemia of various muscle regions. The altered twist may represent the lowest possible energy state for the ventricle as a whole to assume.
It is interesting to hypothesize that the shortening rate for S/P TGA patients at the atrioventricular valve level was significantly different from the shortening rate at the apical short-axis level and from that for Fontan patients because the degree of paradoxical regional wall motion was greater for S/P TGA patients at the septal (posterior) wall than for Fontan patients at the inferior wall at that level (rate for S/P TGA is almost twice that for Fontan). At the apical level, the degrees of paradoxical regional wall motion were approximately equal. It is even more interesting to note that although strain was significantly greater in the S/P TGA group at the apical short axis, this difference was not translated into shortening rate because it was not statistically significant between the two groups at that short-axis level.
Comparison With Previous Studies of Normal LV Twist
Previous studies have suggested that the normal and hypertrophied LVs do not demonstrate the same twist and radial motion characteristics that we have observed in both of our patient subtypes. Maier et al27 found in both control and hypertrophic cardiomyopathy groups a clockwise rotation at the base (5.0±2.4° and 5.0±2.5°, respectively) and a counterclockwise rotation at the apex (−9.6±2.9° and −7.3±5.2°, respectively). These authors failed to track motion throughout systole. Although Ingels et al,25 Arts et al,21 24 and Chadwick20 did not describe the direction of twist in either clockwise or counterclockwise directions in humans and intact dog, they noted that the apical segment rotated in an opposite direction to the basal segment, ie, there was a wringing motion of the LV during contraction. Hansen et al23 found that this torsion between basal and apical segments was decreased during allograft rejection and myocyte necrosis and normalized once the rejection episode was successfully treated. Beyer and Sideman19 noted that LV twist and strain were coupled, and later, Beyer et al17 noted that twist and shortening also were coupled; however, the direction of twist was not defined.
Other investigators have found that rotation was consistently counterclockwise. Buchalter et al26 separated endocardium from epicardium and found that both twisted counterclockwise, with short-axis slices near the base twisting less than those near the apex and endocardial twist>epicardial twist. These authors also failed to track motion throughout systole. Hansen et al22 also found that “counterclockwise twisting about the left ventricular long axis (as viewed from apex to base) accompanied ventricular ejection in all patients” with implanted radiopaque markers (7.6° to 10.7° at the midventricular level and 13.3° to 23.4° at the apical level). Young et al29 confirms that twisting is “anticlockwise.” Researchers at our laboratory also believe this to be the case28 (Fig 3C⇑).
The present study of systemic RVs differs markedly from both views of ventricular twist, strain, and shortening of the normal LVs (Fig 3C⇑). No studies of normal LVs describe simultaneous counterclockwise and clockwise rotation in the same short-axis plane or describe a transition zone. Systemic RVs were found to twist the same way whether at apical or atrioventricular valve (basal) short-axis levels, similar to the findings of Buchalter et al26 (who found the same twist at three short-axis levels). Furthermore, although Beyer et al17 found a positive correlation between normal LV systolic twist and radial shortening, our findings that a transition zone of no twist is present with any one of three types of radial motion—none, inward, or outward (paradoxical)—show that single ventricles uncouple the twist–shortening relation. We did note, however, that superior walls of systemic RVs, which have the greatest counterclockwise motion, also have the greatest radial contraction.
Because magnetic resonance imaging measures fixed planes in space, through-plane motion of the heart may add artifactual deformation, as Moore et al46 pointed out. It should be noted that they described a type of tagging that is different from ours (radial versus orthogonal grid) and compare three-dimensional strains with two-dimensional strains. However, they observed that axial strains in the radial and circumferential dimensions are fairly close when comparing corrected three-dimensional strain with uncorrected two-dimensional strain (<.05 difference). Furthermore, in normal subjects, Hoffman et al47 showed that twist and twist reversal found when tracking through-plane motion are similar in direction and actually enhanced in magnitude from those images that do not track the through-plane motion. Because the present study deals with thick slices relative to the amount of through-plane motion, this is not believed to have an appreciable affect on the findings.
No data are available concerning normal RV twist and radial motion or normal pediatric LV mechanics. Even if these data were available, they might not add more information than a comparison to the normal LV since the normal RV pumps blood against pulmonary vascular resistance. Nevertheless, a comparison would strengthen our study results.
Ages and time from operation for S/P TGA and Fontan patients are significantly different. Because the S/P TGA repair is rarely done today and magnetic resonance imaging was unavailable years ago, patients matched for age and time from operation are hard to find. The present study provides preliminary observations, and we intend to follow these two groups of patients. Although quantitative changes may occur in Fontan patients as they grow, it is unlikely that the fundamental differences in regional strain, twist, and radial motion we have observed between S/P TGA patients, Fontan patients and subjects with a normal LV will change with time.
Finally, because we have no data regarding the compliance of the lungs of these patients during tidal breathing occurring during the scanning, we cannot fully determine the effect this could have on our observations.
V-V interaction, which has been known to occur in the normal human heart, plays a measurable role in the systemic RV. Regional wall motion and strain may play an important role in the long-term energetics of the heart and may be an important factor in the mechanics of ventricular pressure generation. Results from the present study suggest that regional strain, twist, and radial motion are markedly different in the systemic RV, depending on whether an LV is present to augment ventricular function, and differ significantly from those of the normal human LV. Areas of clockwise, counterclockwise, and no twist (transitional zone) occur, and these regions differ between Fontan and S/P TGA patients. Radial shortening was not uniformly distributed around the myocardial wall, with some regions having a greater radial motion (ie, doing more work) than others and some regions moving paradoxically during systole.
The present study of V-V interaction coupled with our previous observations14 15 of altered regional strain across surgical subgroups of the Fontan procedure lays the groundwork from which surgical outcomes may be evaluated in the future.
This work was funded in part by National Institutes of Health grant RO1-HL-29886. Dr Fogel was supported by a fellowship grant of the American Heart Association, Southeastern Pennsylvania Affiliate, and the American Academy of Pediatrics Section of Cardiology. Dr Hoffman was supported as an established investigator of the American Heart Association. We acknowledge the invaluable programming expertise of John Hoford, BS; the engineering and programming acumen of Krishanu B.Gupta, PhD; and our physicists, Andrew Bogdan, PhD, and John Hazelgrove, PhD, who wrote the pulse sequences for the SPAMM images.
- Received October 10, 1994.
- Revision received January 9, 1995.
- Accepted January 17, 1995.
- Copyright © 1995 by American Heart Association
Feneley MP, Gavaghan TP, Baron DW, Branson JA, Roy PR, Morgan JJ. Contribution of left ventricular contraction to the generation of right ventricular systolic pressure in the human heart. Circulation. 1985;71:473-480.
Santamore WP, Lynch PR, Heckman JL, Bove AA, Meier GD. Left ventricular effects on right ventricular developed pressure. J Appl Physiol. 1976;41:925-930.
Brinker JA, Weiss JL, Lappe DL, Rabson JL, Summer WR, Permutt S, Weisfeldt ML. Leftward septal displacement during right ventricular loading in man. Circulation. 1980;61:626-633.
Fontan F, Baudet E. Surgical repair of tricuspid atresia. Thorax. 1971;26:240-248.
Lang P, Norwood WI. Hemodynamic assessment after palliative surgery for hypoplastic left heart syndrome. Circulation. 1983;68:104-108.
Baffa JM, Rychik J, Gullquist SD, Barber G, Jacobs ML, Norwood WI. Outcome following bidirectional cavo-pulmonary anastomosis prior to modified Fontan procedure. J Am Coll Cardiol. 1991;17:33A. Abstract.
Quaegebeur JM, Rohmer J, Brom AG, Tinkelenberg J. Revival of the Senning operation in the treatment of transposition of the great arteries. Thorax. 1977;32:517-524.
Fogel MA, Gupta KB, Weinberg PM, Hoffman EA. Strain and wall motion analysis of single ventricles throughout staged Fontan reconstruction using magnetic resonance tagging. J Am Coll Cardiol. 1993;21:2A. Abstract.
Fogel MA, Gupta KB, Weinberg PM, Hoffman EA. Strain analysis of single ventricles prior to and after staged Fontan procedure using magnetic resonance imaging. J Am Coll Cardiol. 1992;19:233A. Abstract.
Gupta KB, Bogdan AA, Fellows KE, Hoffman EA. Finite strains using SPAMM in normal human left ventricles. Circulation. 1991;84(suppl II):II-159. Abstract.
Beyer R, Yin F, Hausknecht M, Weisfeldt M, Kass D. Dependence of left ventricular twist-radial shortening relations on cardiac cycle phase. Am J Physiol. 1989;257:H1119-H1126.
Beyer R, Sideman S. A computer study of left ventricular performance based on fiber structure, sarcomere dynamics, and transmural electrical propagation velocity. Circ Res. 1984;55:358-375.
Hansen DE, Daughters GT, Alderman EL, Ingels NB, Miller DC. Torsional deformation of the left ventricle midwall in human hearts with intramyocardial markers: regional heterogeneity and sensitivity to inotropic effect of abrupt rate changes. Circ Res. 1988;62:941-952.
Hansen DE, Daughters GT, Alderman EL, Stinson EB, Baldwin LC, Miller DC. Effect of acute human allograft rejection on left ventricular systolic torsion and diastolic recoil measured by intramyocardial markers. Circulation. 1987;76:998-1008.
Ingels NB, Daughters GT, Stinson EB, Alderman EL. Measurements of midwall myocardial dynamics in intact man by radiography of surgically implanted markers. Circulation. 1975;52:859-867.
Buchalter MB, Weiss JL, Rogers WJ, Zerhouni EA, Weisfeldt ML, Beyar R, Shapiro EP. Noninvasive quantification of left ventricular rotational deformation in normal humans using magnetic resonance imaging myocardial tagging. Circulation. 1990;81:1236-1244.
Maier SE, Fischer SE, McKinnon GC, Hess OM, Krayenbuehl HP, Boesiger P. Evaluation of left ventricular segmental wall motion in hypertrophic cardiomyopathy with myocardial tagging. Circulation. 1992;86:1919-1928.
Hoffman EA. Constancy of total heart volume: an imaging approach to cardiac mechanics. In: Sideman S, Beyer R, eds. Imaging Measurements and Analysis of the Heart. New York, NY: Hemisphere Publishing Corporation; 1991:3-19.
Young AA, Imai H, Chang CN, Axel L. Two-dimensional left ventricular deformation during systole using magnetic resonance imaging with spatial modulation of magnetization. Circulation. 1994;89:740-752.
Braunwald E, Sonnenblick EH, Ross J. Mechanisms of cardiac contraction and relaxation. In: Braunwald E, ed. Heart Disease: A Textbook of Cardiovascular Medicine. Philadelphia, Pa: WB Saunders; 1992:371.
Burns JW, Covell JW, Myers R, Ross J. Comparison of directly measured left ventricular wall stress and stress calculated from geometric reference figures. Circ Res. 1971;28:611-621.
Hood WP, Thompson WJ, Rackley C, Rolett EL. Comparison of calculations of left ventricular wall stress in man from thin-walled and thick-walled ellipsoidal models. Circ Res. 1969;24:575-582.
McHale PA, Greenfield JC. Evaluation of several geometric models for estimation of left ventricular circumferential wall stress. Circ Res. 1973,33:303-312.
Waldman LK, Fung YC, Covell JW. Transmural myocardial deformation in the canine left ventricle. Circ Res. 1985;57:152-163.
Rademakers FE, Rogers WJ, Guier WH, Hutchins GM, Siu CO, Weisfeldt ML, Weiss JL, Shapiro EP. Relation of regional cross-fiber shortening to wall thickening in the intact human heart: three-dimensional strain analysis by NMR tagging. Circulation. 1994;89:1174-1182.
De Floriani L. Surface representations based on triangular grids. Visual Comput. 1987;3:27-50.
Lee DT, Schachter BJ. Two algorithms for constructing a Delaunay triangulation. Int J Comput Info Sci. 1980;9:219-242.
Cristesue MG, Ritman EL, Hoffman EA. Mechanics, energetics and constant heart volume: atrial fibrillation vs sinus rhythm. FASEB J. 1990;4:A433. Abstract.
Hoffman EA, Gupta KB, Bogdan AA, Fellows KE. Through plane motion and the assessment of LV short axis mechanics. FASEB J. 1992;6:A1232. Abstract.